Skip to content

dash-libs/dash-ml

Repository files navigation

DashML — Databricks Library

CI PyPI License

Part of the Dashlibs suite — Databricks libraries built for business users.

ML lifecycle management: preprocessing, drift monitoring, evaluation (SHAP + model cards), governance artifacts, and champion/challenger promotion — driven from one notebook UI, backed by Unity Catalog and MLflow.

Installation

%pip install dash-mlops

Quick Start

import dashml
dashml.launch()   # Opens interactive UI in your Databricks notebook

What it covers

Area Entry points
Preprocessing clean_dataframe(), dashml.transforms (outlier removal, binning, lag features, ...)
Drift monitoring ModelMonitor (PSI + chi-squared, optional auto-retrain trigger)
Evaluation explain_features() (SHAP), build_model_card(), check_thresholds()
Governance build_governance_artifacts() (signature, features, fairness, approval record)
Registry RunTracker, register_model(), promote_challenger() (UC @champion alias)
Experimentation dashml.experiment — compare/promote MLflow runs
Serving dashml.serving.sync_serving_endpoint()

Everything beyond ModelMonitor and the notebook UI is also directly importable for use in a training script — launch() is the guided path, not the only path.

Part of Dashlibs

Library Purpose
dash-dq Data Quality
dash-synthetic Synthetic Data Generation
dash-ml ML Lifecycle Management
dash-ingest Data Ingestion
dash-gov Data Governance
dash-ontology Ontology & Lineage for AI

License

Apache 2.0

About

Databricks library — part of the Dashlibs suite

Resources

Stars

Watchers

Forks

Packages

 
 
 

Contributors

Languages